XTDYNGCCE: Stata module to provide bootstrap and generalized inference for CCE panel data models

xtdyngcce provides two estimation commands for panel data models with unobserved common factors (interactive fixed effects), extending the Common Correlated Effects framework of Pesaran (2006) in two directions that are not currently available in Stata. xtdyncce fits the pooled CCE estimator in dynamic panels and implements the recursive-design wild bootstrap of Juodis, Stauskas and Tromp (2026, arXiv:2609.31442). In each replication the dependent variable is regenerated recursively and the cross-sectional averages are re-estimated from the bootstrap data; both steps are required by the theory and distinguish the procedure from a residual wild bootstrap. The command also provides the three bias corrections benchmarked in that paper -- the half-panel jackknife of Dhaene and Jochmans (2015), the analytical correction of Moon and Weidner (2017), and the fixed-T minimum-distance correction of De Vos and Everaert (2021) -- a leave-one-out jackknife variance in which the factor estimates also omit the deleted unit, and an iterated (double) bootstrap for studentized confidence intervals. xtgcce fits the Generalized CCE mean-group estimator of Lu, Su and Ba (2026, Journal of Econometrics 253, 106183). It augments the regression with the time-series averages of the regressors in addition to the usual cross-sectional averages. The time-series averages act as control variables for the factor loadings, which makes the estimator doubly robust: consistent under either the usual CCE rank condition or the paper's control-variable condition, with no efficiency loss when the rank condition holds.